BACKGROUND Previous studies show that rest-redistribution thallium imaging is useful in the assessment of myocardial viability. The impact of such studies on patient outcome is not well defined. This study examined the prognostic value of tomographic rest-redistribution 201T1 imaging in 81 medically treated patients with coronary artery disease and left ventricular dysfunction. RESULTS Rest-redistribution single-photon emission computed tomographic images were obtained and analyzed quantitatively. The segmental thallium uptake (20 segments per patient) was interpreted as normal, reversible defect, mild to moderate fixed defect, or severe fixed defect. The thallium images were abnormal in 80 patients, with no redistribution (no ischemia) in 43 patients and redistribution (ischemia) in 38 patients. The left ventricular ejection fraction was 27% +/- 8% in patients with no redistribution and 26% +/- 7% in patients with redistribution (difference not significant). In patients with no ischemia, there were 7 +/- 5 severe fixed defects and 5 +/- 4 mild to moderate fixed defects per patient. In patients with ischemia there were 7 +/- 4 reversible defects, 3 +/- 3 mild to moderate fixed defects, and 5 +/- 4 severe fixed defects per patient. The number of any abnormal segments was 11 +/- 5 in patients with no ischemia and 14 +/- 4 in patients with ischemia (p = 0.03). During a mean follow-up of 31 +/- 24 months, there were 11 cardiac deaths in patients with no ischemia (26%) and 22 in patients with ischemia (58%); the survival rate was worse in patients with than without ischemia (p < 0.05). Multivariate Cox survival analysis on important clinical, angiographic, and thallium variables showed that the presence of redistribution was an independent predictor of death (x2 = 5; p = 0.03). CONCLUSIONS Patients with left ventricular dysfunction and redistribution on rest thallium imaging, a marker of hibernating myocardium, have a higher mortality rate with medical therapy than do patients with a comparable degree of left ventricular dysfunction but with fixed defects only. Thus observations similar to those made with positron emission tomography can be made in a much more straightforward, simple, and probably cost-effective manner with single-photon emission computed tomography.